Generative AI services such as ChatGPT are increasingly used in everyday and professional contexts, offering functional benefits while also raising various risks. Despite growing interest in the adoption of generative AI, limited attention has been paid to how users’ expectations and usage experiences jointly shape their evaluations. Drawing on Expectation Disconfirmation Theory (EDT), this study examines how perceived utilitarian value and perceived risk of generative AI influence user satisfaction, with a comparative focus on positive and negative disconfirmation groups. Utilitarian value is conceptualized in terms of efficiency, accessibility, and productivity, whereas perceived risk encompasses performance risk, privacy concerns, hallucination, ethical issues, and potential capability loss. Survey data from experienced ChatGPT users were analyzed using partial least squares structural equation modeling with a bootstrapping procedure. The findings show that perceived utilitarian value consistently enhances performance perception and expectation levels across both disconfirmation groups. In contrast, the effects of perceived risk and expectations on user evaluation processes differ depending on disconfirmation type, suggesting that users interpret risks and unmet expectations in distinct ways. Performance perception emerges as a strong determinant of satisfaction in both groups, while the role of expectations varies according to expectation performance disconfirmation. These findings extend EDT to the generative AI context and provide implications for managing user expectations and functional value to support sustainable AI usage.
Park et al. (2026) studied this question.